Applied Ergonomics 59 (2017) 387e400

Contents lists available at ScienceDirect

Applied Ergonomics
journal homepage: www.elsevier.com/locate/apergo

A simplified thermoregulation model of the human body in warm

conditions
Baizhan Li a, b, Yu Yang a, b, c, Runming Yao a, b, c, *, Hong Liu a, b, Yongqiang Li a, b, c
a
Joint International Laboratory of Green Buildings and Built Environments, Ministry of Education, Chongqing University, Chongqing, China
b
National Centre for International Research of Low-carbon and Green Buildings, Ministry of Science and Technology, Chongqing University, Chongqing,
China
c
School of the Built Environment, University of Reading, UK

a r t i c l e i n f o a b s t r a c t

Article history: Thermoregulation models of the human body have been widely used in thermal comfort studies. The
Received 20 January 2016 existing models are complicated and not fully verified for application in China. This paper presents a
Received in revised form simplified thermoregulation model which has been statistically validated by the predicted and measured
23 September 2016
mean skin temperature in warm environments, including 21 typical conditions with 400 Chinese sub-
Accepted 26 September 2016
Available online 12 October 2016
jects. This model comprises three parts: i) the physical model; ii) the controlled system; and iii) the
controlling system, and considers three key questions formerly ignored by the existing models including:
a) the evaporation efficiency of regulatory sweat; b) the proportional relation of total skin blood flow and
Keywords:
Thermoregulation model
total heat loss by regulatory sweating against body surface area; and c) discrepancies in the mean skin
Thermal response temperatures by gender. The developed model has been validated to be within the 95% confidence in-
Skin temperature terval of the population mean skin temperature in three cases.
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction The simplified Gagge's 2-node model of thermoregulation

The thermal interaction of the human body with the environ-
ment involves two processes: i) the heat transfer between the
human body and the thermal environment, simultaneously
including radiation, convection, conduction, evaporation and
respiration; and ii) the self-regulation function of the human body
which responds to varied thermal environments, such as vaso-
constriction, vasodilation, shivering and sweating (Cheng et al.,
2012). Thermoregulation models of the human body are devel-
oped to simulate these two processes of interaction and predict the
human thermal response under different thermal conditions and
have been widely used in the field of physiology or thermal comfort
studies (Parsons, 2014). An accurate thermoregulation model will
help improve the accuracy of the current thermal comfort predic-
tion models, and provide a basic theoretical analysis of the accuracy
of the various models in application (De Giuli et al., 2014;
Holopainen et al., 2014).
* Corresponding author. Joint International Laboratory of Green Buildings and
Built Environments, Ministry of Education, Chongqing University, Chongqing,
China. School of the Built Environment, University of Reading, UK.
E-mail addresses: r.yao@reading.ac.uk, r.yao@cqu.edu.cn (R. Yao).
(Gagge et al., 1971) is one of the most popular models in the field of
thermal comfort study. Moreover, various complex thermoregula-
tion models have been further developed by improving the
modelling of body segmentation, particularly for heat insulation
(Arezes et al., 2013), thermoregulatory systems and heat transfer
(Fiala et al., 2001; Munir et al., 2009; Stolwijk, 1971; Werner and
Webb, 1993; Xu and Werner, 1997), considering individual body
characteristics (Takada et al., 2009; Zhang et al., 2001), and
increasing the number of body segments to obtain a higher reso-
lution temperature distribution on the skin surface (Huizenga et al.,
2001; Tanabe et al., 2002).
These models are mostly developed based on European or
American populations; however, their accuracy lacks effective
validation (Yang et al., 2015a). There is little strong evidence in the
existing research to show that existing models are applicable to the
Chinese population. Thermal comfort prediction for the Chinese
people still remains in an early research stage which is largely
based on the modification of the traditional models but is still
lacking systematic analysis (Zhou et al., 2013, 2014). In this context,
this paper aims to i) validate the predictive accuracy of the classic
Two-Node model for the Chinese population; and ii) develop and
validate a new simplified model based on the laboratory

http://dx.doi.org/10.1016/j.apergo.2016.09.010
0003-6870/© 2016 Elsevier Ltd. All rights reserved.
388 B. Li et al. / Applied Ergonomics 59 (2017) 387e400
experiments.
The mean skin temperature was used for the validation of the
developed model. In the existing studies, skin temperature has
been demonstrated to be strongly related to the thermal interaction
between the human body and the thermal environment, which is
also an important indicator of thermal comfort (Parsons, 2014). It
has been successfully used to validate increasingly complex and
sophisticated predictive models for thermoregulatory responses,
and to build thermal sensation models.
The systems predicting the interaction between people and
their environment are complex (Andrew Thatcher, 2016). Here, the
developed model shows advantages over many other existing
models. The individual differences in human thermal responses are
caused by some characteristics which can be quantitatively defined
(age, height, weight, etc.), but may also contains some of the po-
tential differences which are not so easily described such as the
property of each layer of the body including core composition,
muscle composition, fat composition and skin composition
respectively. The mean basal metabolic rate of the Chinese popu-
lation is re-measured in this study. It has allowed the simplification
of the human body abstraction as a cylinder with its specific geo-
metric dimensions and heat transfer direction, which cannot be
provided by simply adjusting the parameters of existing models for
the Chinese population. Meanwhile, the introduction of a cylinder
model and the development of control plates make it more
convenient and accurate in application compared to other models
(Yang et al., 2015b).
2. Description of the new model
The proposed model consists of three parts: the physical model
of the human body, the controlled system and the controlling
system.
2.1. Physical model of the human body
In this physical model, the human body is abstracted as a
Fig. 1. The schematic diagram of the physical model of the human body.
cylinder consisting of four concentric layers: the core, muscle, fat
and skin. A central pool of blood delivers the arterial blood to the
capillaries and tissues in each layer, and meantime the blood flows
back to the central pool through the veins. The schematic diagram
of this physical model is shown in Fig. 1. Assuming that the physical
characteristics in each layer are uniform, the physical parameters of
each layer are recalculated from the data of reference (Gordon et al.,
1976; Stolwijk, 1971) and listed in Table 1.
Considering the size of the physical model, the height dimen-
sion is far greater than the radius dimension. In the simulation, heat
is only supposed to be transferred in a radial direction. Radial de-
pendency of temperature is calculated in the model. In this paper,
abbreviations with subscripts of i ¼ 1,2,3,4 represent the layers of
core, muscle, fat and skin respectively. The subscripts b and cl
represent the central blood and clothing nodes respectively.
The geometric characteristics of the physical model can be
calculated from the basic information of the human body (gender,
height, weight and body fat percentage). The surface area A (m2) of
a Chinese human body can be obtained by Equations (1) and (2) for
male and female subjects (Wang, 1994). The length of the cylinder L
(m) and the external radius of layer i (which is denoted by rs,i(m))
can be calculated by Equations (5) and (6) respectively.
A ¼ 0:0057H þ 0:0121W þ 0:0882 ðfor malesÞ (1)
A ¼ 0:0073H þ 0:0127W  0:2106 ðfor femalesÞ (2)
mi ¼ am;i W (3)
Vi ¼ mi =ri (4)
, !
X
4
L ¼ A2 4p Vi (5)
i¼1
 B. Li et al. / Applied Ergonomics 59 (2017) 387e400 389

Table 1
The physical parameters of the layers (Gordon et al., 1976; Stolwijk, 1971).

Layer Core Muscle Fat Skin

Density (kg/m3) r1 ¼ 977 r2 ¼ 1115 r3 ¼ 850 r4 ¼ 1000

Specific heat (J/(kg  C)) c1 ¼ 2968 c2 ¼ 3105 c3 ¼ 2510 c4 ¼ 3760
Heat conductivity coefficient (w/(m k)) l1 ¼ 0.42 l2 ¼ 0.66 l3 ¼ 0.21 l4 ¼ 0.21

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Based on the physical model of the human body, the scheme of
P4 thermal interaction for humans with the environment can be seen
i¼1 Vi
rs;i ¼ (6) from Fig. 2.
pL

Where H is the height (cm); W is the weight (kg); am,i is the pro-
portion of layer i in the weight, which is recalculated by reference to 2.2.1. Energy equation
Stolwijk (1971) and shown in Table 2. The energy equation in a one-dimensional cylindrical coordi-
nate system based on classical heat transfer theory is:
Vi is the volume of layer i (m3). !
vT v2 Ti 1 vTi
ri ci i ¼ li þ þ Mb;i þ Mw;i þ Ms;i
Therefore, gender, height, weight and body fat percentage can vt vr 2 r vr
be used as the inputs for the physical model. The default values for þ bQi cb rb ðTb  Ti Þ (7)
Chinese male and female subjects in the model are set as 170 cm,
70 kg, 20% and 160 cm, 55 kg, 25% respectively (Yang, 2015). Where, Ti(r,t) is the temperature of tissue ( C); Mb,i(r) is the heat
production by basic metabolism in Watts per cubic metre (W/m3);
Mw,i(r,t) is the heat production by activity metabolism per cubic
2.2. Controlled system
metre (W/m3); Ms,i(r,t) is the heat production by shivering per cubic
metre (W/m3); b is the counter-current factor, by which the
The controlled system is used to simulate the internal heat
approximate heat exchange between arterial blood and venous
transfer of the body and the heat transfer between the body surface
blood is considered. The effect of the counter-current usually takes
and the thermal environment.
place in cold conditions and b equals 1 in this paper for warm
conditions. Qi(r,t) is the blood flow per cubic metre [m3/(s m3)]; cb is
Table 2 the specific heat of central blood, which equals 3760 [J/(kg  C)]; rb is
The proportions of layers by weight (Stolwijk, 1971). the density of central blood, which equals 1000 (kg/m3); Tb is the
Core (am,1) Muscle and fat (am,2þam,3) Skin (am,4) Total
temperature of central blood ( C).Mb,i can be calculated from the
total basal metabolic rate Mb (W/m2), which is related to gender
22% 73% 5% 100%
and age; the data for the Chinese population is shown in Table 3

Fig. 2. The scheme for the thermal interaction of humans with the environment.
390 B. Li et al. / Applied Ergonomics 59 (2017) 387e400

Table 3 from respiration per body surface area respectively (W/m2) (Fanger,
The mean basal metabolic rate of the Chinese population Mb (W/m2) (Yao, 2005). 1970):
Gender Age

11e15 16e17 18e19 20e30 31e40 41e50 Over 51 Cres ¼ 0:0014Mð34  Ta Þ (15)
Male 54.28 53.69 46.14 43.82 44.05 42.77 41.38
Female 47.88 50.44 42.77 40.68 40.79 39.52 38.47 Eres ¼ 0:0000173Mð5867  Pa Þ (16)
The continuity of temperature and heat flux at an interface
(Yao, 2005): between two layers of different tissues is expressed in Equations
(17) and (18).
Mb;i ¼ di Mb A=Vi (8)
   
Ti rs;i  0; t ¼ Tiþ1 rs;i þ 0; t ði ¼ 1; 2; 3Þ (17)
Where, di is the proportion of the basal metabolic rate taken up by
layer i.
   
Gordon et al. (1976) gave the reference value as shown in vTi vTiþ1
li ¼ liþ1 ði ¼ 1; 2; 3Þ (18)
Table 4. vr rs;i 0 vr rs;i þ0
Assuming Mw,i is only related to the activity level and this energy
is produced by the muscle, the following equations are used to At the skin surface, the heat brought to the surface by conduc-
describe Mw,i: tion from the deep body is equal to the heat removed from the
surface by evaporation and conduction. Therefore, the boundary
Mw;1 ¼ Mw;3 ¼ Mw;4 ¼ 0 (9) condition at the skin surface is:

Mw;2 ¼ ðM  Mb ÞA=V2 (10) vTi T  Tcl  

li ¼ fcl i  Esw  Ediff i ¼ 4; r ¼ rs;4 (19)
vr 0:155Icl
Where, M is the metabolic heat generation for a certain activity (W/
m2), which is provided by the international standard (ASHRAE-55, Where, Icl(t) is the total thermal insulation of clothing (clo), which
2004); Mb is assumed to be the basal metabolic rate for the popu- can be estimated or calculated by ISO 7730 (ISO-7730, 2005); Esw(t)
lation aged from 20 to 30. is the evaporative heat loss by sweating per body surface area (W/
m2); Ediff(t) is the diffusion evaporative heat loss per body surface
Ms,i and Qi are variable and controlled by the controlling system. area (W/m2).
Tb is considered to be independent of the radius and determined fcl(t) is the clothing area factor, which can be roughly estimated
by the energy equation: by Equation (20) (McCullough et al., 1985):

Zrs;i fcl ¼ 1:0 þ 0:3Icl (20)

dTb X 4  
rb cb Qb ¼ brb cb Qi Tv;i  Tb (11)
dt i¼1 Esw is regulated by the thermoregulatory controlling system; Ediff
rs;i1
can be calculated from Equation (21) (Fanger, 1970).
Where, rs,0 represents the radius of the cylinder's centre, which is    
equal to 0; Tv,i(r,t) is the venous temperature ( C), and it is assumed Ediff ðtÞ ¼ 0:00305 256T rs;4 ; t  3373  Pa (21)
to be equal to the temperature of the adjacent tissue, that is:
Considering the fact that people are usually in light clothing in
Tv;i ðr; tÞ ¼ Ti ðr; tÞ (12) warm conditions, the clothing node is simplified as follows:

Qb is the total blood flow to the cylinder (m3) and is obtained by dTcl T  Tcl
mcl ccl ¼A i  AðC þ RÞ (22)
integration over the volume of a cylinder with length L: dt 0:155Icl

X
4 Zrs;i Where, mcl is the weight of the clothing (kg), the default value is set
Qb ¼ 2pL Qi rdr (13) as 0.2 kg in warm conditions; ccl is the specific heat of clothing (J/
i¼1 rs;i1
(kg  C)), Yi et al. (2004) provide the data for some common mate-
rials, the value for cotton is 1210.
C(t) and R(t) are the convective and radiative heat losses from
the outer surface of a clothed body, both of which are related to the
2.2.2. Boundary and initial conditions difference between the mean temperature of the outer surface of
The boundary condition at the centre of the cylinder is: the clothed body Tcl ( C) and the mean air temperature Ta ( C), as
shown in Equation (23).
vTi
jr ¼ 0 ¼ Cres þ Eres (14)
vr C ¼ fcl hc ðTcl  Ta Þ (23)
Where, Cres and Eres are the convective and evaporative heat loss Where, hc is the convective heat transfer coefficient (W/m2· C);
Equations for estimating hc are expressed as (Fanger, 1970):
Table 4
The proportion of the basal metabolic rate for each layer of the human body (Gordon  pffiffiffiffiffiffi 
et al., 1976). hc ¼ 2:38ðTcl  Ta Þ0:25 when 2:38ðTcl  Ta Þ0:25 > 12:1 Va
Core (d1) Muscle (d2) Fat (d3) Skin (d4) Total (24)
4.1% 74.9% 0.7% 20.9% 100.0%
 B. Li et al. / Applied Ergonomics 59 (2017) 387e400 391

pffiffiffiffiffiffi  pffiffiffiffiffiffi  2.3.1. Signal input

hc ¼ 12:1 Va when 12:1 Va  2:38ðTcl  Ta Þ0:25 (25) The input for the controlling system consists of two signals
collected from the core layer and skin layer respectively. These
signals are integrated temperatures formed by the hypothalamus
Ar h i
R ¼ fcl εs ðTcl þ 273Þ4  ðTr þ 273Þ4 (26) temperature and the skin temperature, which can be expressed as:
A
Where,    
Errc ¼ T rs;0  Tset;c þ Rd  T 0 rs;0 (31)
ε ¼ average emissivity of clothing or body surface,
(dimensionless).    
Errs ¼ T rs;4  Tset;s þ Rd  T 0 rs;4 (32)
s ¼ Stefan-Boltzmann constant, 5:67  108 W/(m2 K4).
Ar ¼ effective radiation area of body, m2. Where, Errc are the input signals of the core layer ( C) and Errs are
Tr ¼ mean radiant temperature, C the input signals of skin layer ( C); T(rs,0) represents the tempera-
ture of the thermoreceptor in the hypothalamus and T(rs,4) repre-
The ratio Ar/A is 0.70 for a sitting person and 0.73 for a standing sents the temperature of the thermoreceptor under the skin; Tset,0
person (Fanger, 1970). Emissivity ε is close to unity (typically 0.95), and Tset,s are the temperature set points for the core and skin
unless special reflective materials are used or high-temperature respectively ( C); Rd is the dynamic sensitivity of the thermore-
sources are involved. ceptor and T0 represents the temperature change rate ( C/s).
The mean radiant temperature Tr can be determined by the The set point temperatures (Tset,c andTset,s) can be acquired by
measurement of the black globe temperature (Tg) and the air simulating the body under a thermally neutral condition with no
temperature and air velocity at the level of this globe (ISO-7726, work and no regulatory control. The experiment was designed to
2001). show the body's adaptation in a neutral thermal environment. The
1) In the case of natural convection: human body was in a steady state at the end of the exposure stage
"   #1=4 when the change rate of the body's average skin temperature was
 4 0:25  108 Tg  Ta 1=4   <0.01  C/min. The steady neutral environmental parameters and
Tr ¼ Tg þ 273 þ  Tg  Ta
εg D human body temperature set points are recorded in Tables 5 and 6.
Parameters for male and female subjects to achieve thermal
 273
neutrality were obtained by the previous experimental study and
(27) these two sets are listed in Table 5. The calculated set point tem-
peratures are different for each gender and results are shown in
Where, εg is the emissivity of the black globe (dimensionless); D is Table 6.
the diameter of the globe (m). The value or the quantitative analysis for Rd is not totally
For the standard globe D ¼ 0.15 m, εg¼0.95 (matt black paint) and revealed according to the existing references (Kobayashi and
Equation (27) becomes: Tanabe, 2013; Tanabe et al., 2002), it represents the human sensi-
h 4 1=4  i1=4 tivity to the change of the ambient temperature. Here we chose
Tr ¼ Tg þ 273 þ 0:4  108 Tg  Ta  Tg  Ta  273 empirical values for the prediction accuracy of the model: when
T0 >0, Rd ¼ 0(0); when T0 <0, Rd ¼ 0 (s) for males, and Rd ¼ 1800 (s) for
(28) females (Yang, 2015).
2) In the case of forced convection:
" #1=4
 4 1:1  108  Va0:6   2.3.1.1. Vasomotion. The total blood flow for layers BFi (l/h) is
Tr ¼ Tg þ 273 þ Tg  Ta  273
εg  D0:4 calculated by Equations (33)e(36). The blood flow for core and fat
tissue remains constant. For muscle compartments, a blood flow of
(29)
1.0 l/h was required for 1.16 W metabolic heat production (Tanabe
For the standard globe: et al., 2002). The skin blood flow is controlled by the effect of
vasodilation or vasoconstriction in the thermoregulation system,
h 4  i1=4 which is assumed to be proportional to the body surface A.
Tr ¼ Tg þ 273 þ 2:5  108  Va0:6 Tg  Ta  273 (30)

The initial conditions which specify the values of all dependent BF1 ¼ 255 (33)
variables at time zero should be provided. The initial values may be
equilibrium values which are obtained from a previous steady-state
calculation (denoted as Ta j, Tr j, va j, RHa j, Icl j, Mj), or they may be
non-equilibrium values which result from a previous transient Table 5
Parameters for a thermally neutral condition.
process. In either case, they consist of body temperature specifi-
cations at the instant the transient process begins, which are Parameters Ta( C) RH(%) Va(m/s) Tg( C) M(W/m2) Icl(clo)
denoted as Ti j. Male 26 74.11 0.05 26.47 58.15 0.4
Female 26.2 73.55 0.06 26.42 55.15 0.4

2.3. Controlling system

Table 6
The controlling system is used to simulate the thermoregulatory The set point temperatures of the human body.
control mechanisms in the human body. It includes the regulation
Temperature set points ( C) Tset,c Tset,s
of blood flow, sweating and shivering. The proposed controlling
system is based on the traditional controlling system provided in Male 36.94 34.16
Female 36.67 33.8
Stolwijk (1971) and further improved by an empirical formula.
392 B. Li et al. / Applied Ergonomics 59 (2017) 387e400

 
BF2 ¼ 14:74 þ Mw;2 þ Mshi  A=1:16 (34) Mshi ¼ 24:4Errc Errs =A ðif Errc > 0 or Errs > 0; then Mshi ¼ 0Þ (42)

BF3 ¼ 3:5 (35)

2.4. Numerical solution and computer programming
11:89 þ DL A
BF4 ¼  2Errs =10  (36)
1 þ ST 1:89 The numerical solution for solving the thermoregulation model
is an explicit method with centred finite difference. The spatial grid
DL ¼ 117Errc þ 7:5Errs ðif DL < 0; then DL ¼ 0Þ (37) is radially divided with the spacingDr ¼ 0.002 m. The time incre-
ment is set as 1s.
ST ¼ 0:63Errc  0:63Errs ðif ST < 0; thenST ¼ 0Þ (38) The simulation program is written in Matlab 2010 and the
programming flow chart is shown in Fig. 3. It consists of five parts:
Therefore: INPUT, PHYSICAL, CONTROLLING, CONTROLLED and OUTPUT. Their
functions are shown Table 7 below:
Qi ¼ 0:278  106 BFi =Vi (39)
3. Validation of the models

A model is by definition simpler than the system it attempts to

2.3.1.2. Sweating. The heat loss by regulatory sweating per body represent. Thus, validation by comparing the model prediction with
surface area Esw (W/m2) is calculated by Equations (40 and (41): the experimental data from human subjects is necessary. Not only
For males, does this process identify the prediction accuracy of the model, but
. it also helps to define the range of conditions in which the model is
Esw ¼ ð223Errc þ 20Errs Þ2Errs =10 1:89 (40) applicable.
Skin temperature is strongly related to the thermal interaction
For females, between the human body and the thermal environment (Gagge
. et al., 1971; Stolwijk, 1971), which is also an important indicator
Esw ¼ ð111Errc þ 10Errs Þ2Errs =10 1:89 (41) of thermal comfort (Cheng et al., 2012; Fanger, 1970). It has been
successfully used to validate increasingly complex and sophisti-
Esw shows differences according to gender, the value for females is cated predictive models for thermoregulatory responses (Munir
smaller than that for males under the same conditions. The total et al., 2009; Yi et al., 2004; Zolfaghari and Maerefat, 2010), and to
heat loss by regulatory sweating is also proportional to the body build thermal sensation models (Lomas et al., 2003; Wang, 1994).
surface area A. Therefore, in this paper we choose the mean skin temperature to
validate the accuracy of the models.
The method for the model validation can be found in Yang et al.
2.3.1.3. Shivering heat production. The shivering heat production (2015a). According to this approach, the accuracy of the model has
per body surface area (W/m2) is calculated by Equation (42): been primarily examined by inferential statistical analysis, and then

Fig. 3. Flow chart of the simulation program. (n: calculated numbers; t: calculated time; Dt:time step; te: simulation time).
 B. Li et al. / Applied Ergonomics 59 (2017) 387e400 393

Table 7 experiments with a total of 400 subjects. For comparison purposes,

Model structure distribution. the existing Two-Node model, which is regarded as one of the most
Model Definition and coverage classic simplified thermoregulation models in thermal comfort
structure studies, has also been validated by comparing its predictions with
Human body Build physical model of human body and calculate the the experimental data of Case 1.
model geometric, physiological and physical parameters of the model.
Controlling Calculate the parameters of the controlling system based on the
system human thermal physiological responses and thermal
3.1. Case 1
adaptability.
Controlled Calculate the body temperature according to the equations in a
system controlled system based on the application of classical heat A temperature step-changing experiment was carried out to
transfer theory between the human body and the dynamic validate the performance of the models under transient conditions.
thermal environment. Ten male and ten female healthy subjects were recruited randomly
Input Environment parameters
parameters Mean air temperature, Mean radiant temperature, Mean air
to participate in the experiment. During the experiment, all the
velocity, Relative humidity, Metabolic rate, Thermal insulation subjects were required to wear uniform clothing including light
of clothing. long-sleeve cotton shirts and trousers, and light shoes with a total
Human body parameters clothing insulation level of 0.4clo (1clo equals 0.155 m2 K/W). Five
Gender, age, height, weight, body fat percentage.
environment conditions were designed in this experiment. In each
Output Body temperature in any transient of the thermal process.
parameters condition, the subjects firstly experienced a step-change thermal
process from a neutral environment (Environment I) to a typical
warm environment (Environment II), and then stayed in Environ-
assessed through the Bland-Altman method (Martin Bland & ment II for 1800 s. The subjects then returned to the neutral con-
Altman, 1986) if necessary. By the validation, the accuracy of the dition (Environment III) for another 1800 s. The basic information
model can be classified into three levels: IdThe model's prediction of the subjects and the environment conditions in Case 1 are listed
is sufficiently statistically accurate; IIdThe model's prediction is in Tables 8 and 9 respectively.
sufficiently accurate to be used in applications; IIIdThe model's During the experiment, skin temperature measurements at 13
prediction is not sufficiently accurate. locations on the body including the forehead, chest, back, upper
Three series of experimental data from human subjects under arm (right and left), lower arm (right and left), dorsal hand (right
the typical warm conditions have been obtained, they are: and left), calf (right and left), and thigh (right and left) were per-
formed automatically with a frequency of 0.5 Hz. The 8-point
Case 1: clothed subjects in step-changing environments weighted method (Gagge and Nishi, 2011) was adopted to calcu-
(26  Ce28 C/29 C/30 C/32 C/34 
C, then back to 26  C). late the body mean skin temperature (Tsk) as represented by
Case 2: nude subjects in step-changing environments Equation (43):
(28  Ce32 C/35  C).
Case 3: clothed subjects in typical warm, steady state, Tsk ¼ 0:07Tforehead þ 0:175Tchest þ 0:175Tback þ 0:07Tupper arm
environments. þ 0:07Tlower arm þ 0:05Thand þ 0:19Tthigh þ 0:020Tcalf

The proposed model has been validated by all the three sets of (43)
In order to validate and compare the performance of the new
Table 8 model and the classic Two-Node Gagge Model (Gagge et al., 1971),
Subjects' information in Case 1 (mean ± standard deviation). the two models were operated to simulate the above thermal
Subjects Male Female process. The predicted and measured skin temperatures during the
3600 s in the experiments of Case 1 are shown in Fig. 4 and Fig. 5 for
Age 24 ± 1 24 ± 1
Height (m) 170 ± 7 159 ± 6 male and female subjects respectively. According to the model
Weight (kg) 58 ± 5 51 ± 8 evaluation method (Yang et al., 2015a), statistical validation is
Body fat percentage (%) 16.9 ± 2.97 25.6 ± 5.3 conducted by comparing model predictions with the confidence
Clothing insulation (clo) 0.4 ± 0 0.4 ± 0 intervals of the population means. For the model, all the predictions
Activity level (met) 1.0 ± 0 1.0 ± 0
lie within the 95% confidence interval of the population means,

Table 9
Thermal conditions of the experiment in Case 1 (mean ± standard deviation).

Case 1 Air temperature ( C) Relative humidity (%) Air velocity (m/s) Globe temperature ( C)

Condition 1 Environment I 26.0 ± 0.2 71.9 ± 3.0 0.05 ± 0.00 26.4 ± 0.2
Environment II 28.2 ± 0.1 54.6 ± 0.2 0.18 ± 0.04 28.4 ± 0.1
Environment III 26.0 ± 0.2 70.5 ± 3 0.06 ± 0.01 26.5 ± 0.2
Condition 2 Environment I 26.1 ± 0.1 77.3 ± 2.8 0.05 ± 0.01 26.5 ± 0.1
Environment II 29.0 ± 0.1 54.7 ± 0.8 0.17 ± 0.07 29.3 ± 0.1
Environment III 26.2 ± 0.2 75.9 ± 2.5 0.05 ± 0.01 26.6 ± 0.1
Condition 3 Environment I 26.1 ± 0.2 71.5 ± 3.9 0.06 ± 0.01 26.5 ± 0.3
Environment II 30.3 ± 0.2 58.1 ± 3.3 0.14 ± 0.09 30.4 ± 0.2
Environment III 26.0 ± 0.3 71.0 ± 3.6 0.06 ± 0.02 26.5 ± 0.3
Condition 4 Environment I 26.5 ± 0.3 70.8 ± 3.7 0.18 ± 0.04 26.5 ± 0.3
Environment II 32.0 ± 0.1 53.9 ± 3.9 0.20 ± 0.03 32.0 ± 0.1
Environment III 26.8 ± 0.3 69.6 ± 4.9 0.08 ± 0.02 26.8 ± 0.3
Condition 5 Environment I 26.0 ± 0.1 77.2 ± 1.1 0.05 ± 0.01 26.3 ± 0.2
Environment II 33.8 ± 0.1 56.1 ± 0.7 0.20 ± 0.04 33.9 ± 0.0
Environment III 26.2 ± 0.1 75.4 ± 2.5 0.06 ± 0.01 26.5 ± 0.1
394 B. Li et al. / Applied Ergonomics 59 (2017) 387e400
Fig. 4. Model validation for Case 1 (male subjects).
indicating no statistically significant difference between the pop-
ulation means and the predictions of the model. The accuracy of the
new model in predicting the transient responses for subjects with
light clothing is evaluated as Level I. However, for the Gagge Model,
significant differences were found between the model predictions
and the measured data, especially in high-temperature conditions
(Conditions 4 and 5). The Gagge Model cannot be statistically
validated in this case. Further empirical validation for the Gagge
Model is carried out by the Bland-Altman method (Martin Bland &
Altman, 1986; Yang et al., 2015a). The 'limit of agreement' of the
predictions of the Gagge Model and population means are calcu-
lated as [-0.49, 0.09] and [-0.52, 0.19] for males and females
respectively, which suggests the predictions of the Gagge Model
may be 0.49  C below or 0.09  C above the measured sample means
for male subjects (similarly 0.52  C below or 0.19  C above for fe-
male subjects). The Gagge Model is unacceptable if we regard the
accuracy requirement for the skin temperature in thermal comfort
studies as ‘the difference between the model prediction and the
sample mean in most cases must be less than 0.3  C’, and it should
be evaluated as Level III. Therefore, predictions from our new model
are superior to the Gagge model in case 1.
3.2. Case 2
A further experiment was set up for nude subjects to validate
the new simplified model. Ten half-naked healthy male students
were randomly recruited as subjects and each participated in one of
two condition sets in this experiment. In both conditions, the
subjects experienced a temperature step-change process from a
neutral environment (Environment I) to a typical warm environ-
ment (Environment II) and then stayed in Environment II for a
period of 3600 s. The data for the subjects and the thermal condi-
tions in Case 2 are listed in Tables 10 and 11 respectively. Skin
temperatures were collected as in Case 1.
The predicted values of the new model and experimental results
of body skin temperature are shown in Fig. 6. For all the conditions,
no significant difference was found between the model prediction
and the target population. According to the model validation
method (Yang et al., 2015a), the accuracy of the model is evaluated
as Level I, which means that the new model is statistically accurate
to simulate the transient mean skin temperature for the nude
population under typical warm conditions.
3.3. Case 3
Steady-state experiments were conducted to validate the new
model's accuracy in various typical warm environments. The warm
environments include the variation of temperature, humidity and
air velocity. Three series of human exposure experiments with
fourteen conditions in total were carried out in a climate chamber.
In each series, ten male and ten female healthy subjects were
recruited. During the experiment, subjects were required to wear
uniform clothing including short-sleeve shirts, shorts and light-
weight shoes with an insulation level of 0.26clo (1clo equal to
0.155 m2 K/W). In each of the conditions, subjects were given
sedentary office activities and 120 min exposure was provided for
subjects to reach a steady state. At the end of the exposure, skin
temperatures at 13 locations on the body were recorded as illus-
trated in Case 1. The data on the subjects and the thermal condi-
tions in Case 3 are listed in Tables 12 and 13 respectively. The
default body fat percentages are set as 15% and 25% for male and
female subjects respectively.
The predicted values of the new model and experimental results
of the body skin temperature under different conditions are shown
in Fig. 7 and Fig. 8 for male and female subjects respectively. The
skin temperature for both males and females can be accurately
predicted by the developed model. The performance of the new
model in the steady state condition is evaluated as Level I.
4. Discussion
A thermoregulation model of the human body has been pre-
sented. The new model is simplified compared to most existing
models (e.g. (Fiala et al., 2001; Huizenga et al., 2001; Wan and Fan,
 B. Li et al. / Applied Ergonomics 59 (2017) 387e400 395

Fig. 5. Model validation for Case 1 (female subjects).

Table 10 more practical in application and yet further optimization of the

Subjects' information in Case 2 (mean ± standard deviation). empirical parameters can be more easily achieved.
Age 24 ± 1 In contrast with the existing models, the major contribution of
Height (m) 174 ± 6 this newly developed model is to point out three questions
Weight (kg) 60 ± 7 neglected in thermoregulation modelling and optimize them. They
Body fat percentage (%) 15.8 ± 2.5
Clothing insulation (clo) 0.03 ± 0
involve a) the evaporation efficiency of regulatory sweat; b) the
Activity level (met) 1.0 ± 0 effect of body surface area on thermoregulatory calculations; c) the
gender difference. The developed model establishes its advantages
over current models as shown in Table 15.
2008). etc.). Nevertheless, the model is statistically accurate in
estimating the transient skin temperature of the human body as 4.1. Evaporation efficiency of regulatory sweating
discussed in the next section. The model has been validated by
comparing its simulation with experimental results from responses In the new model, the calculation of heat loss by regulatory
by a total of 400 subjects under typical warm conditions, as shown sweating is optimized based on the empirical formula in Stolwijk
in Table 14. The advantage of this simplified model is that it can be (1971). The value of Esw calculated from the new model is

Table 11
Thermal conditions of the experiment in Case 2 (mean ± standard deviation).

Case 2 Air temperature ( C) Relative humidity (%) Air velocity (m/s) Globe temperature ( C)

Condition 1 Environment I 28.0 ± 0.2 61.7 ± 5.2 0.06 ± 0.01 28.1 ± 0.3
Environment II 31.9 ± 0.1 55.1 ± 0.3 0.09 ± 0.02 32.0 ± 0.1
Condition 2 Environment I 28.2 ± 0.1 60.4 ± 2.8 0.06 ± 0.01 28.3 ± 0.2
Environment II 34.7 ± 0.1 55.8 ± 0.3 0.18 ± 0.04 34.7 ± 0.1
396 B. Li et al. / Applied Ergonomics 59 (2017) 387e400

Fig. 6. Validation of the developed model for case 2.

Table 12
Subjects' information in Case 3 (mean ± standard deviation).

Series A B C

Subjects Male Female Male Female Male Female

Age 24 ± 1 24 ± 1 23 ± 1 24 ± 1 23 ± 1 24 ± 1
Height (m) 173 ± 5 160 ± 7 175 ± 5 159 ± 4 172 ± 6 160 ± 3
Weight (kg) 61 ± 4 50 ± 6 67 ± 10 46 ± 4 63 ± 9 48 ± 5
Clothing insulation (clo) 0.26 ± 0 0.26 ± 0 0.26 ± 0 0.26 ± 0 0.26 ± 0 0.26 ± 0
Activity level (met) 1.2 ± 0 1.2 ± 0 1.2 ± 0 1.2 ± 0 1.2 ± 0 1.2 ± 0

obviously lower than that obtained from the model in Stolwijk
(1971) under the same conditions. The original empirical formula
for calculating Esw (Stolwijk, 1971) was based on a common
assumption that all the sweating can be evaporated directly.
However, the reality is that considerable sweating will adhere to
the body surface, clothing or drop down from the body rather than
evaporating instantly as heat loss. Thus, the original calculation for
Esw overestimates the evaporation heat loss and modified empirical
formulas considering the ‘evaporation efficiency’ are employed as
Equation (40) and Equation (41). In order to illustrate the optimi-
zation of evaporation by sweating, the study applied the original
and modified equations for Esw respectively to simulate male sub-
jects' skin temperature in the thermal process of Condition 5 in
Case 1, the results for the initial 1800 s are shown in Fig. 9. It is
Table 13
Thermal conditions of the experiment in Case 3 (mean ± standard deviation).

Case 3 Ambient temperature ( C) Relative Velocity(m/s) Black-bulb

Series Conditions

A 1 26.9 ± 0.2 54 ± 4 0.11 ± 0.02 26.6 ± 0.1

2 28.9 ± 0.2 55 ± 7 0.11 ± 0.04 28.5 ± 0.2
3 31.0 ± 0.2 51 ± 7 0.14 ± 0.04 30.4 ± 0.1
4 32.9 ± 0.2 54 ± 5 0.12 ± 0.02 32.3 ± 0.1
B 1 25.6 ± 0.1 41 ± 1 0.08 ± 0.05 25.6 ± 0.1
2 25.9 ± 0.1 60 ± 1 0.1 ± 0.06 25.6 ± 0.1
3 28.0 ± 0.1 40 ± 2 0.07 ± 0.01 27.6 ± 0.1
4 27.9 ± 0.1 60 ± 1 0.09 ± 0.03 27.6 ± 0.2
5 29.8 ± 0.1 42 ± 2 0.1 ± 0.02 29.4 ± 0.2
6 29.9 ± 0.1 60 ± 1 0.09 ± 0.03 29.4 ± 0.1
C 1 28.0 ± 0.1 90 ± 1 0.6 ± 0.03 28.0 ± 0.1
2 28.1 ± 0.2 90 ± 1 0.79 ± 0.04 28.0 ± 0.2
3 30.0 ± 0.2 80 ± 1 0.81 ± 0.04 29.8 ± 0.2
4 32.0 ± 0.2 80 ± 1 0.79 ± 0.03 31.9 ± 0.2

obvious that when applying the original calculation for Esw, the
predicted skin temperature is significantly lower than the
measured value, which is the result of the overestimation of Esw.
This optimization of Esw has also been validated in all the high
temperature conditions in our experiments. As the assumption that
the sweating can be evaporated away completely is quite common
in the existing models (e.g. (Gagge et al., 1971; Munir et al., 2009;
Tanabe et al., 2002).), the question of sweating evaporation effi-
ciency should be worthy of attention in further models.

4.2. Effect of body surface area on thermoregulatory calculation

The body surface area (A) decides the heat transfer at the skin
surface thereby further influencing the body temperature. How-
ever, in most existing models (e.g. (Gagge et al., 1971; Stolwijk,
1971; Tanabe et al., 2002).), the effect of A is ignored when calcu-
lating the value of total heat loss by regulatory sweating or total
skin blood flow, both of which should be theoretically proportional
to A. As a modification, the new model introduces the variable A
into the formula when estimating the total heat loss by regulatory
sweating and total skin blood flow. Fig. 10 shows the results of the
model with (Equation (36) and Equation (43)) or without (Equation
(44) and Equation (45)) consideration of A respectively when pre-
dicting the first 1800 s of Condition 5 in Case 1 for females. Ac-
cording to the model evaluation method, the model's performance
is improved from Level II to Level I by introducing A into the
calculation of total heat loss by regulatory sweating and total skin
blood flow. The effect of A on the simulation results is usually
insignificant for populations with similar physiques; however,
when populations have significantly different values of A, e.g. the
value is calculated as 1.6 m2 for Chinese females compared with the
1.89 m2 (Stolwijk, 1971) for an American or European ‘standard
man’, the effect of A should not be ignored. This optimization, based
on the effect of A, takes into account the different physiques of
populations, which makes the model more reasonable and reliable.

11:89 þ DL
BF4 ¼  2Errs =10 (44)
1 þ ST
.
Esw ¼ ð111Errc þ 10Errs Þ2Errs =10 A (45)

4.3. Gender difference

In the new model, the difference in gender is introduced into

calculations of basal metabolic rate, fat percentage, body surface
Fig. 7. Validation of the Developed Model for Case 3 (male subjects) (The error bars
and evaporation heat loss etc., all of which will affect in the final
represent the 95% confidence interval for the population means).
398 B. Li et al. / Applied Ergonomics 59 (2017) 387e400

Table 15
Advantages and development of the new model over existing models.

Advantages Development compared with the existing models

Differences between 1 Physical parameters of the physical abstract

races model.
2 Metabolic heat production in the controlled
system.
3 Parameter selection and optimization in the
controlling system.
Differences between 1 Height, weight, body fat rate.
individuals 2 Age Factor.
3 Gender factor.
Simplicity and accuracy 1 Simplicity: physical structure, the input
parameters.
2 Accuracy: model validation, accuracy level.

Fig. 9. Example to illustrate the optimization of evaporation by sweating.

Fig. 10. Example to illustrate the effect of body surface area on thermoregulatory
calculations.
Fig. 8. Validation of the Developed Model for Case 3 (female subjects) (The error bars
represent the 95% confidence interval for the population means).

experimental results can be well illustrated by the new model with

temperature distribution of the body. The experiment in Case 1 the consideration of gender difference, which is an improvement
shows that under the same thermal conditions (e.g. Condition 5), on the existing models.
the characteristics of temperature regulation are significantly The differences in the thermal responses of males and females
different for male and female subjects. As shown in Fig. 11, these have been noticed by many studies because of their different

Table 14
Accuracy level in Case validation.

Case study Working conditions The accuracy of the model

Case 1 Transient response for subjects with light clothing Level I

Case 2 Transient response for the nude population Level I
Case 3 Steady state Level I
 B. Li et al. / Applied Ergonomics 59 (2017) 387e400 399

thank the China Scholarship Council (201306050050,

201406050066) for the sponsorship for a one-year academic
visiting study at the University of Reading.

Nomenclature

A body surface area (m2)

Ar effective radiation area of body (m2)
BFi Total blood flow of layer i (l/h)
c specific heat [J/(kg  C)]
C convective heat losses (W/m2)
Cres respiratory convective heat flow per body surface area
(W/m2)
DL vasodilation signal (l/h)
Ediff diffusion evaporative heat loss per body surface area (W/
Fig. 11. Example to illustrate the gender difference.
m2 )
Eres evaporative heat loss from respiration per body surface
physiological characteristics. The discriminations have been made area (W/m2)
in this study such as the calculation of body surface area “A” and the Errc input signals of core layer ( C)
mean basal metabolic rate. Besides, the gender factor has a direct Errs input signals of skin layer ( C)
influence on the physical model of the human body including the Esw evaporative heat loss by sweating per body surface area
height, weight and body fat percentage which cannot be avoided. (W/m2)
So this study tries to solve the prediction deviation caused by fcl clothing area factor
gender by deducing a separate formula for males and females, H height of human body (cm)
together with certain actual measured data for the two groups. hc convective heat transfer coefficient (W/m2$ C)
Icl total thermal insulation of clothing (clo)
L length of cylinder (m)
5. Conclusions M total metabolic rate (W/m2)
Mb total basal metabolic rate (W/m2)
This paper presents a simplified thermoregulation model of the Mb,i heat production by basic metabolism per cubic metre (W/
human body. The new model is statistically accurate in predicting m3 )
mean skin temperature in warm environments and has been mcl weight of the clothing (kg)
verified by three sets of experiments including 21 typical condi- Ms,i heat production by shivering per cubic metre (W/m3)
tions with 400 subjects in total. This model consists of three parts: Mshi shivering heat production (W/m2)
i) the physical model, which is an abstraction of the real body; ii) Mw,i heat production by activity metabolism per cubic metre
the controlled system, which is used to simulate the heat transfer of (W/m3)
the body and environment; and iii) the controlling system, which Pa water vapour partial pressure (Pa)
describes the thermoregulatory control mechanisms of the human Qb total blood flow to cylinder (m3)
body. Qi blood flow per cubic metre of layer i [m3/(s m3)]
This newly-developed model has been used to analyse the mean r radius of cylinder (m)
skin temperate based on the data from the group of subjects and R radiative heat losses (W/m2)
reveals three key phenomena which are normally ignored in the Rd dynamic sensitivity of thermoreceptors (s)
existing models: a) the role played by the evaporation efficiency of RH Relative humidity (%)
regulatory sweating; b) the proportional relation of total skin blood rs,i external radius of layer I (m)
flow and total heat loss by regulatory sweating against body surface ST vasoconstriction signal (l/h)
area and c) there are discrepancies in the mean skin temperature t time (s)
between the genders. The newly-developed model has been sub- T0 temperature change rate ( C/s)
ject to experimental validation which supports its optimizational Ta mean air temperature ( C)
modifications that lead to advantages in application accuracy Tb temperature of blood ( C)
compared to current mainstream models. Meanwhile, the intro- Tcl mean temperature of the outer surface of the clothed
duction of a cylinder model and the development of control plates body ( C)
make it more convenient in application compared to other human Ti temperature of tissue ( C)
thermal prediction models. Tr mean radiant temperature ( C)
This model can be widely applied in the field of thermal comfort Tset,c core temperature set point ( C)
study, in particular for the prediction of transient skin temperature Tset,s skin temperature set point ( C)
in different thermal processes. It is applicable for the iterative Tv venous temperature ( C)
calculation of environmental and physiological parameters in the Va mean air velocity (m/s)
dynamic thermal environments in terms of different times and Vi volume of layer I (m3)
spaces. W weight of human body (kg)
am,i proportion of layer i in weight (%)
Acknowledgement b counter-current factor
ε average emissivity of clothing or body surface
The authors would like to thank the Natural Science foundation l heat conductivity coefficient [W/(m$k)]
(No. 51561135002) and 111 Project (No. B13041) for their financial r density (kg/m3)
support on the research. Yu Yang and Yongqiang Li would like to s Stefan-Boltzmann constant[W/(m2$K4)]
400 B. Li et al. / Applied Ergonomics 59 (2017) 387e400
Subscript
i layer number
1 layer of core
2 layer of muscle
3 layer of fat
4 layer of skin
b central blood node
cl clothing node
References
Andrew Thatcher, P.H.P.Y., 2016. Human factors for a sustainable future. Appl. Ergon.
57, 1e7.
Arezes, P., Neves, M.M., Teixeira, S., Le~ ao, C.P., Cunha, J., 2013. Testing thermal
comfort of trekking boots: an objective and subjective evaluation. Appl. Ergon.
44, 557e565.
ASHRAE-55, 2004. Thermal Environmental Conditions for Human Occupancy.
American Society of Heating, Refrigerating and Air-conditioning Engineers.
Cheng, Y., Niu, J., Gao, N., 2012. Thermal comfort models: a review and numerical
investigation. Build. Environ. 47, 13e22.
De Giuli, V., Zecchin, R., Corain, L., Salmaso, L., 2014. Measured and perceived
environmental comfort: field monitoring in an Italian school. Appl. Ergon. 45,
1035e1047.
Fanger, P.O., 1970. Thermal Comfort. Analysis and Applications in Environmental
Engineering. Thermal Comfort. Analysis and Applications in Environmental
Engineering.
Fiala, D., Lomas, K.J., Stohrer, M., 2001. Computer prediction of human thermoreg-
ulatory and temperature responses to a wide range of environmental condi-
tions. Int. J. Biometeorol. 45, 143e159.
Gagge, A.P., J.A.J., S., Y., N., 1971. An effective temperature scale based on a single
model of human physiological temperature response. ASHRAE Trans. 77,
247e262.
Gagge, A.P., Nishi, Y., 2011. Heat exchange between human skin surface and thermal
environment. Compr. Physiol.
Gordon, R.G., Roemer, R.B., Horvath, S.M., 1976. A mathematical model of the human
temperature regulatory system-transient cold exposure response. Biomedical
Engineering. IEEE Trans. 434e444.
Holopainen, R., Tuomaala, P., Hernandez, P., H€ akkinen, T., Piira, K., Piippo, J., 2014.
Comfort assessment in the context of sustainable buildings: comparison of
simplified and detailed human thermal sensation methods. Build. Environ. 71,
60e70.
Huizenga, C., Hui, Z., Arens, E., 2001. A model of human physiology and comfort for
assessing complex thermal environments. Build. Environ. 36, 691e699.
ISO-7726, 2001. Ergonomics of the Thermal Environment. Instruments for
Measuring Physical Quantities International Organization for Standardization.
ISO-7730, 2005. Ergonomics of the Thermal Environment, Analytical Determination
and Interpretation of Thermal Comfort Using Calculation of the PMV and PPD
Indices and Local Thermal Comfort Criteria. International Organization for
Standardization.
Kobayashi, Y., Tanabe, S.-i, 2013. Development of JOS-2 human thermoregulation
model with detailed vascular system. Build. Environ. 66, 1e10.
Lomas, K., Fiala, D., Stohrer, M., 2003. First Principles Modeling of Thermal Sensa-
tion Responses in Steady-state and Transient Conditions.
Martin Bland, J., Altman, D., 1986. Statistical methods for assessing agreement be-
tween two methods of clinical measurement. Lancet 327, 307e310.
McCullough, E.A., Jones, B.W., Huck, J., 1985. A comprehensive data base for esti-
mating clothing insulation. Ashrae Trans. 91, 29e47.
Munir, A., Takada, S., Matsushita, T., 2009. Re-evaluation of Stolwijk's 25-node
human thermal model under thermal-transient conditions: prediction of skin
temperature in low-activity conditions. Build. Environ. 44, 1777e1787.
Parsons, K., 2014. Human Thermal Environments: the Effects of Hot, Moderate, and
Cold Environments on Human Health, Comfort, and Performance. Crc Press.
Stolwijk, J., 1971. A Mathematical Model of Physiological Temperature Regulation in
Man. National Aeronautics and Space Administration.
Takada, S., Kobayashi, H., Matsushita, T., 2009. Thermal model of human body fitted
with individual characteristics of body temperature regulation. Build. Environ.
44, 463e470.
Tanabe, S.-i., Kobayashi, K., Nakano, J., Ozeki, Y., Konishi, M., 2002. Evaluation of
thermal comfort using combined multi-node thermoregulation (65MN) and
radiation models and computational fluid dynamics (CFD). Energy Build. 34,
637e646.
Wan, X., Fan, J., 2008. A transient thermal model of the human body-
eclothingeenvironment system. J. Therm. Biol. 33, 87e97.
Wang, X., 1994. Thermal Comfort and Sensation under Transient Conditions.
Department of Energy Technology, Division of Heating and Ventilation, the
Royal Institute of Technology.
Werner, J., Webb, P., 1993. A six-cylinder model of human thermoregulation for
general use on personal computers. Ann. physiol. Anthropol. 12, 123.
Xu, X., Werner, J., 1997. A dynamic model of the human/clothing/environment-
system. Applied human science. J. physiol. Anthropol. 16, 61e75.
Yang, Y., 2015. Indoor Thermal Response of Human Body in Uniform Environment
(Warm Conditions). Phd. Thesis. Chongqing University.
Yang, Y., Yao, R., Li, B., Liu, H., Jiang, L., 2015a. A method of evaluating the accuracy of
human body thermoregulation models. Build. Environ.
Yang, Y., Yao, R., Li, B., Liu, H., Jiang, L., 2015b. A method of evaluating the accuracy of
human body thermoregulation models. Build. Environ. 87, 1e9.
Yao, T., 2005. Physiology. Fudan University Press, Shanghai.
Yi, L., Fengzhi, L., Yingxi, L., Zhongxuan, L., 2004. An integrated model for simulating
interactive thermal processes in humaneclothing system. J. Therm. Biol. 29,
567e575.
Zhang, H., Huizenga, C., Arens, E., Yu, T., 2001. Considering individual physiological
differences in a human thermal model. J. Therm. Biol. 26, 401e408.
Zhou, X., Lian, Z., Lan, L., 2013. An individualized human thermoregulation model
for Chinese adults. Build. Environ. 70, 257e265.
Zhou, X., Zhang, H., Lian, Z., Zhang, Y., 2014. A model for predicting thermal
sensation of Chinese people. Build. Environ. 82, 237e246.
Zolfaghari, A., Maerefat, M., 2010. A new simplified thermoregulatory bioheat
model for evaluating thermal response of the human body to transient envi-
ronments. Build. Environ. 45, 2068e2076.